A low rating is not one kind of review

Among the reviews carriers left on CDLScan, 94.8% carry the lowest rating. That figure describes those reviews, not drivers in general: carriers write after a bad run, so the share tells you what shows up once someone bothers to report it. Almost every write-up in this corpus is already a complaint. The useful question is not “was the rating bad.” It is “bad how.”

The remaining 5.2% — ratings above the floor — is a thin slice. I will not build a hiring philosophy on it. I will use it the way the data allows: as a contrast that shows the lowest rating is not a single story.

What changes when the rating is not the floor

Abandoned-truck mentions are in 33.5% of all reviews, 34.2% of the lowest-rating reviews, and 21.0% of the rest. That is a 13.2-point drop once you leave the floor. The event does not vanish. It gets less common.

Paid travel with no arrival goes from 6.3% on the floor to 1.0% above it. No-show goes from 5.6% to 1.4%. Going unreachable from 8.3% to 3.4%. Walking off a loaded trailer from 5.7% to 2.4%. Theft or money games from 9.2% to 3.0%. Drugs or alcohol from 12.1% to 3.8%. Those are operational walk-aways and onboarding losses, and they concentrate where the carrier reached for the worst score.

What barely moves is also useful. Late or missed deliveries sit at 9.8% on the floor and 9.6% above it. Accident or damage: 13.8% versus 12.9%. Attitude: 13.9% versus 12.6%. Performance: 3.7% versus 3.2%. A late load or a dented fender can produce a review that is still angry without being the worst number on the scale. A driver who left the truck usually cannot.

Verdict language — the author’s “do not bother” recommendation, not a fact about the driver — drops hard: 33.4% on the floor, 9.3% above it. That is the carrier deciding the next employer should not waste a call. It is attitude toward the reader, not a second incident.

One small move goes the other way, and I will not overclaim it: a stated personal or family reason is 2.2% on the floor and 2.9% above it. Quit-early is 0.7% versus 1.0%. Those categories are tiny. The direction is interesting. It is not a finding you can take to a safety meeting.

How these shares are counted is on the methodology page. The 5.2% slice is small. Treat every contrast as a difference in mix, not as a portrait of “good” reviews. There are almost no good reviews here.

Why this matters on a desk

If you treat every lowest rating as “they abandoned a truck,” you will misread the file. A third of those reviews mention abandonment. That means most of them do not. The same floor also holds late loads, damage, attitude, and a lot of verdict-only write-ups where the carrier is warning you without giving you a clean event.

If you treat a 2 or a 3 as a pass, you will also misread the file. Abandoned-truck mentions are still in 21.0% of that thin slice. A rating above the floor is not a character reference. It is a slightly different mix of complaints.

The pattern I’d watch for is the event, not the star. Left the truck, never showed, ticket unused, stopped answering — those sentences do the work. The number next to them tells you how hot the author was, not what happened.

Typical sequence, built from the mix, not from one person’s file: you open a profile, see the floor, and stop reading. Or you see a 3 and skip it. In both cases you skipped the sentence that would have changed the hire. The rating was never the screen.

What official reports do with this

Nothing. PSP, an MVR, and the Clearinghouse do not store carrier star ratings. They store filed crashes, inspections, license actions, program events. A lowest-rating review about a late delivery and a lowest-rating review about a truck left at a stop look the same to those systems: they look like nothing.

That is why the mix inside the floor matters. The packet you already buy is silent on both. The write-up is not. Read the event. Use the rating only as heat.

The climb in abandonment — 18.8% of reviews in 2018 to 40.4% in 2025, last complete year — sits almost entirely inside this lowest-rating world. I stop that line at 2025. The companion essay on what the 94.8% floor means for the whole site is what it means that almost every review carries the lowest rating.

What to do with a rated file

Open the write-up. Ask what happened before you ask what it scored. If the sentence is left the truck, treat it as a walk-off screen. If the sentence is never showed after a ticket, treat it as an onboarding screen. If the sentence is late and rude, you are in a different conversation — still useful, not the same spend.

Do not average the stars and call it a score. 94.8% of this corpus is already the floor. An average here is a measurement of how often carriers bother to write, not of how often drivers fail.

Then look the name up across write-ups, not across ratings. Check it on CDLScan before you spend. It takes less than a minute. The useful part is the sentence, not the number beside it.

Frequently asked questions

Does 94.8% mean almost every driver is rated a 1?

No. 94.8% is the share of reviews, among those carriers left on CDLScan, that carry the lowest rating. Carriers write after a bad run. It is a fact about the corpus, not a rate for all CDL holders.

If most reviews are already a 1, why look at ratings at all?

Because the 5.2% above the floor has a different mix. Walk-offs and no-shows concentrate on the floor. Late loads and damage barely move. The star is heat. The sentence is the event.

Can I trust a 3-star review more than a 1?

Not as a pass. Abandoned-truck mentions are still in 21.0% of reviews above the floor. Read the event. Do not upgrade a complaint because it is not the worst number.

Is the 5.2% slice big enough to quote?

It is big enough to show a contrast. It is not big enough to build a second product on. I am using it as a foil for the floor, not as its own population study.

Do official reports use these ratings?

No. PSP, MVR, and the Clearinghouse do not store carrier stars. They store filings. The rating lives only in the review.

What should I tell a hiring manager who wants a score?

Give them the event and the year, not an average star. An average in a 94.8% floor is a statistic about complaining, not about the driver pool.

Where do I read the write-ups?

On CDLScan, before you hire. Look at what happened. Use the rating only to know how done the last carrier was.